Description and retrieval of geometric patterns on surface meshes using an edge-based LBP approach. (October 2018)

• Record Type: Journal Article
• Authors: Moscoso Thompson, Elia; Biasotti, Silvia
• Is Part Of: Pattern recognition. Volume 82(2018:Oct.)
• Page Start: 1
• Page End: 15
• Publication Date: 2018-10
• Subjects: LBP extension -- surface tessellations -- 3D pattern retrieval -- 3D pattern classification; 68U05 -- 65D18; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2018.04.028
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: The edgeLBP extends the LBP technique to surface tessellations, whose faces are convex polygons. The edgeLBP description is robust to non-uniform surface samplings. The edgeLBP aims at retrieving and classifying surfaces according to their geometric patterns. Abstract: While texture analysis is largely addressed for images, the comparison of the geometric reliefs on surfaces embedded in the 3D space is still an open challenge. Starting from the Local Binary Pattern (LBP) description originally defined for images, we introduce the edge-Local Binary Pattern (edgeLBP) as a local description able to capture the evolution of repeated, geometric patterns on surface meshes. Our extension is independent of the surface representation, indeed the edgeLBP is able to deal with surface tessellations characterized by non-uniform vertex distributions and different types of faces, such as triangles, quadrangles and, in general, convex polygons. Besides the desirable robustness properties the edgeLBP exhibits over a number of examples, we show how this description performs well for 3D pattern retrieval and compare our performances with the participants to a recent 3D pattern retrieval and classification contest [1].